Quality assurance of paediatric lateral chest radiographs

Abstract

Lateral chest radiographs are widely used, particularly in low and middle-income countries, for diagnosing pulmonary tuberculous lymphadenopathy. The World Health Organisation recommends that “good-quality chest radiographs (including lateral view, if and where possible) are essential for proper evaluation” of paediatric pulmonary tuberculosis. However, the quality of such radiographs has received little attention and there is no currently available quality assurance tool for evaluating these in children (or adults). AIM: To develop a quality assurance tick-sheet for paediatric lateral chest radiographs and to test it for quantifying radiographic errors. METHOD: A quantitative retrospective cross-sectional study, using a sample of 300 chest radiographs of paediatric patients being investigated for pulmonary tuberculosis for the period February 2008 to January 2013. A quality assurance tick-sheet and scoring system was created, using existing guidelines and reasonable radiological principles. Three readers applied the tick-sheet to the database of radiographs, to establish common errors affecting radiograph quality (radiographer-related and patient-related) and to determine the reliability of each criterion. A majority decision for the presence of each error and an average of the score was reported. Reliability of each quality assurance criterion was evaluated using raw agreement and the kappa statistic.

RESULTS: Of the 300 radiographs, 67 (22.3%) were considered optimal using the tick-sheet’s scoring system i.e. had an average score of less than 1, and 182 (60.7%) demonstrated an average n score of 1-2. Patient-related errors were present in 208 (69.3%) of lateral radiographs, the most prevalent being left-to-right rotation in 153 (51%), under-inspired lungs in 82 (27.3%) and the presence of motion artefact in 27 (9%). Radiographer-related errors were present in 110 (36.7%), the most common of which were under-collimation 59 (19.7%) and over-exposure i.e. too dark 40 (13.3%). For 6 of the 11 criteria there was moderate, fair or substantial agreement (based on raw agreement or kappa): No side marker, Static artifact, Over-exposure i.e. too dark, Under-exposure i.e. too light, Under-collimation and Rotation:left-to-right. For the remaining 5 criteria there was slight agreement or less than chance agreement (based on raw agreement or kappa): Over-collimation, Rotation:sideto- side, Under-inspiration (described for the first time in this paper), Position and Motion artifact. CONCLUSIONS: Where the lateral chest radiograph is often key in diagnosing TB in children and diagnostic doubt resulting from the quality of the radiograph must be minimised, availability and use of the proposed quality assurance tick-sheet will foster more precise radiographic technique and also improve the radiological service offered. 5 of the 11 criteria with poor inter-rater agreement require improvement of the definition of the criterion to increase reliability.